Each influenza hemagglutinin (HA) is a mosaic of conserved and strain-specific epitopes, some of which are recognized by neutralizing antibodies. An influenza vaccine that preferentially elicits neutralizing antibodies that recognize conserved epitopes will be more broadly protective than those now in use. Understanding how the exposure history of subjects, differences in influenza vaccine antigens, and the addition of adjuvants bias the immune response towards different HA epitopes will enable us to design optimal vaccines. The following three hypotheses will be tested. (1] The B-cell repertoires elicited by non-replicating influenza vaccines and by infection differ in degree of polyclonal activation and breadth of neutralization. The repertoires elicited by adjuvanted and non-adjuvanted vaccines differ because adjuvant broadens the range of recognized epitopes. (2] Broadly reactive antibodies, including those recognizing the heterosubtypic stem epitope, are less frequent after true primary than after secondary influenza immunization due to broadening of the response by multiple HA stimulations. (3) Efficient induction of antibodies against a desired epitope requires: (a] proliferation of a favorable germline antibody and (b) an affinity maturation pathway to a desired final specificity. There are preferred germline precursors and maturation pathways for antibodies targeting particular epitopes. Hypotheses 1 and 2 will be tested by mapping the epitopes recognized by the repertoires of human subjects with different exposure histories to influenza infection and/or immunization with adjuvanted or un-adjuvanted vaccines. We will seek broadly neutralizing antibodies (those we want to elicit) to understand their germline precursors and maturation pathways. We will also map the epitopes of the full range of antibodies that bind HA to determine what the humoral immune system